3D printers - fun with FDM printers (Klipper-based)

[Martin Wynne]

.
This topic will be about using the latest high-speed FDM printers based on the Klipper G-code interpreter and driver, for printing timbering bases (and filing jigs). Initially about the Elegoo Neptune 4 series of printers.

For discussions about Marlin-based FDM printers, see instead this topic:

https://85a.uk/templot/club/index.php?threads/3d-printers-fun-with-fdm-printers-marlin-based.277/

and:

https://85a.uk/templot/club/index.php?threads/3d-printer-kingroon-kp3s-pro-s1.850/

and other previous topics about FDM printing.

 

[Martin Wynne]

@gavin @James Walters

Having assembled and adjusted the Neptune 4, I thought I should first try a test piece of timbering base using the printer straight out of the box. From previous experience of FDM printers I wasn't expecting doing that to be very successful.

In the event the result was excellent and amazingly fast. A 6-timber base which was taking about 66 minutes to print on the Kingroon was printed in just 15 minutes, i.e. a speed increase of more than 4-fold. Or a reduction from 11 minutes per timber to only 2.5 minutes per timber. That was using the recommended speed, and there is scope to go even faster.

Needless to say the out-of-the-box settings had some issues which will need dealing with:

1. it printed an unnecessary brim around the part. I changed the setting to omit it.

2. the top of the timbers had the usual mottled 3D-printed surface. I switched on the ironing function to produce a smooth surface.

3. the standard setting is for a 10% fill density. I increased it to 40% to match previous timbering bases.

That was using the "Normal" profile with 0.16mm layers.

After making these changes the next print time increased slightly to 16 minutes. This is mainly caused by the ironing function.

4. the soleplate detail was missing. This will likely need some changes to the default thickness dimensions in Templot to better match the layer height.

5. the usual scrub with a stiff nail brush caused the dropper-wire ridges to detach from the base. This will similarly require some changes to the dimensions to better match the layer height.

But there was no stringing at all, the sockets were perfectly clear even at the 220degC nozzle temperature. The sockets were spot-on for size, and the chairs clipped into place very positively.

---

The Neptune 4 printer is supplied with a customized version 4.8 of Cura. It was necessary to use that rather than the 5.6 version which I have been using, because only that has the required printer settings for the Klipper interpreter. I'm intending to get everything copied and set up in Cura 5.6 so that I can use the same version for all. I will post the files when I have them working.

I went with the automatic levelling on the tin PEI bed. It worked ok, but I haven't yet found how to switch off the fade-out in Klipper for constant thickness timbers. On the Neptune 3 I have managed to use a glass bed with perimeter priming run, so I will see if I want to do that with the Neptune 4 -- it does add some faff to the start-up of every print.

The printing starts off quite sedately for the first layer. It then sets off at startling speed for the remainder of the print, and the cooling fans come on full blast. They are very noisy, and this is one downside of the fast printing. It would be worth constructing a soundproof enclosure if the printer is installed in living areas.

Direct computer control of the Neptune 4 uses an Ethernet LAN cable instead of a USB cable on the Neptune 3. This means it can't be controlled directly from within Cura. Instead it is all done from the web browser (Firefox for me). This adds so much additional information and controls that it's probably worth the extra trouble of having to drag and drop a saved G-code file from Cura:

Printer start-up from switch-on is surprisingly slow, and the browser won't connect until the printer screen says it is ready.

Alternatively you can put the G-code file from Cura on a USB memory stick, and run the printer remotely without needing a computer anywhere near.

I will write more when I've done some more tests and adjustments.

cheers,

Martin.

 

[James Walters]

Hi Martin,
This is very interesting indeed. I've all but given up with my FDM machine so am looking for another. I was going to opt for the Kingroon having seen the quality of the prints you have been getting from it, but the Neptune 4 certainly looks like a useful contender.
I shall be following your progress with great interest.

 

[Martin Wynne]

Hi Martin,
This is very interesting indeed. I've all but given up with my FDM machine so am looking for another. I was going to opt for the Kingroon having seen the quality of the prints you have been getting from it, but the Neptune 4 certainly looks like a useful contender.
I shall be following your progress with great interest.

@James Walters @Steve_Cornford @Hayfield @gavin

Hi James,

I'm still very fond of my two Kingroon printers. If you are not in too great a hurry for your prints the clean accurate results from the Kingroon extruder are very appealing. Of the two, the smaller one is actually easier to set up and get going. But 180mm x180mm is undeniably a bit limiting for 4mm scale:

https://altwaylab.co.uk/collections...ingroon-kp3s-3-0-3d-printer-with-meanwell-psu

https://altwaylab.co.uk/collections...ucts/official-kingroon-kp3s-pro-s1-3d-printer

But I became aware that the suggestions and files I've been posting might not be applicable to the newer Klipper-based fast printers, and the only way to find out was to get one and see. I don't want to be saying things on here which are misleading for anyone.

With the recent offers on the Neptune 4 you do get a lot of extra machine for the price difference, although the Amazon Prime price has jumped back up in the 3 days since I ordered one, and the difference for non Amazon Prime members almost pays for Prime for a year:

https://www.amazon.co.uk/dp/B0C9SZ7DCR

The extra features on the Neptune 4 Pro (segmented bed heater, metal rollers, metal top frame) hardly seemed enough to justify its increased cost over the plain Neptune 4.

Kingroon do have a Klipper-based variant, but it's not currently available in the UK via Altway, and the cost of it makes the Neptune 4 a better buy.

There are lots of other Klipper-based printers, but the Neptune 4 seems to be one of the least expensive.

No doubt I will be writing more about it in due course. But not today because I have a hospital appointment this morning.

cheers,

Martin.

 

[Phil G]

Hi Martin,
I too am following your klipper experiments with great interest.
I do get the idea of keeping up with developments for plug track input reasons, and the fact FDM printers are moving to faster prints, Something I can understand.

I have to say though I find the idea of a huge and noisy fan bank with the sole purpose of cooling the filament down quicker, so you can run faster. Somewhat counter intuitive especially when the total height of the job is only 3~4 mm above
the heated bed.
In reality just for plug track timber/sleepers, is it not a case of putting energy in to heat the bed, and then effectively putting more energy in to cool the filament and the bed with it? If so the extra speed gain must come at a higher operating cost.

I guess if you building up models that start to push the Z build volume, it starts to make a lot more sense.

It does strikes me the sweet spot for an FDM printer used mainly for making plug track bases has to be in the 200x200 to 235x 235 bed ranger, with a reliable and trouble free operation as possible.

I would be very keen after your investigation to know, if you think Klipper based faster machines, with all the (extra features you get) meet that requirement?
cheers
Phil

 

[Martin Wynne]

I have to say though I find the idea of a huge and noisy fan bank with the sole purpose of cooling the filament down quicker, so you can run faster. Somewhat counter intuitive especially when the total height of the job is only 3~4 mm above the heated bed.

@Phil G @gavin @James Walters

Hi Phil,

My thoughts too. There is an on-off switch on the top of the fan bank. I have tried printing 4mm/ft scale timbering bases with the extra fans switched off, with no apparent effect on the result. It certainly makes a significant difference to the noise level, although the printer is still quite noisy with the extruder fans and the power supply fan running.

The extra fan bank may be needed for the filing jigs, and for thicker timbering bases in 7mm/ft and larger scales. There is a lot of experimenting still to do.

Elegoo supply a customized version of outdated Cura 4.8 with their printers, but they do now have an updated customized Cura 5.6 version available for download:

https://www.elegoo.com/en-gb/pages/download

The bad news is that this refuses to run alongside the regular Cura 5.6 version, with each blaming the other for configuration errors.

I want to continue using my existing Cura 5.6 so that I can use my current settings for other printers, and get regular Cura updates. I think I shall end up transferring the Elegoo settings manually from one to the other, before I can post any suggested profiles.

I've restored the soleplate and the dropper-wire ridges, but I'm still tinkering with the settings to find the optimum results.

And I mustn't let all this distract me from getting 244a released. There is lots to do.

I would be very keen after your investigation to know, if you think Klipper-based faster machines, with all the (extra features you get) meet that requirement?

I can't say anything about machines I haven't got. But I hope to find out whether the extra cost of the Neptune 4 over other printers is worth it, at least for plug track purposes.

cheers,

Martin.

 

[gavin]

Hello Martin, James and others

Experiments with the Neptune 4plus continue over here with some success, but also a few challenges. However I have no doubt it will work out fine - just need to understand the limitations, which is taking a while as I am only doing it occasionally between work on News 237.

I am very much concentrating on being able to print longer lengths - my test is 300mm long. Seems sensible as the purpose of having the larger bed is to enable this.

I seem to have the tops of the sleepers at an acceptable finish now, but now started having problems with the ends of the lengths sometimes curling up part way through the print - I am assuming this is something to do with adhesion on the plate. Whether I have caused this issue by other settings I have not yet worked out.

So I have this time gone back to printing with the Brim (test 6) and will let you know.

The current settings I have changed from the standard profile are:

Global:
Layer Height = 0.12mm
Build Plate adhesion = Brim
Extruder:
Top Surface Skin Layers = 2
Enable Ironing = True
Iron Only Highest Layer = True
Monotonic Ironing Order = True
Ironing Flow = 15%
Infill Density = 80%

No doubt other people have used different settings.

If the picture uploads correctly, here is my curling template corner part way through a print! Off to the 'day job' now so may try post the result of test 6 later.

Gavin
PS Must work out how to get notified for @gavin!

 

[Martin Wynne]

I seem to have the tops of the sleepers at an acceptable finish now, but now started having problems with the ends of the lengths sometimes curling up part way through the print - I am assuming this is something to do with adhesion on the plate. Whether I have caused this issue by other settings I have not yet worked out.

@gavin @Phil G

Hi Gavin,

A few points:

1. did you give the bed plate a good clean before first use?

2. I can't see much elephant's foot in that photo. To get good adhesion on the bed you need to drop the Z-offset about an extra 0.1mm to get a good squidge against the bed on the first layer. That leaves a slight elephant's foot which is no problem for the track bases -- the connector clips have a rebate around the underside to allow for it.

3. on the PEI beds, they need to be at least 60degC to stick. They lose grip below 60degC and the part falls free at about 40degC. For the track bases there is no need to have the extra-blast fans switched on, which may be reducing the bed temperature -- the thermal sensor is a couple of layers below the bed surface.

4. have you got the rollers on the vertical columns set too tight, preventing the gravity backlash-correction working on Z? You should be able to lift the X-bar a fraction by hand and it drops back easily under its own weight. As supplied the brass nuts should be slightly loose in the frames. This doesn't matter too much with manual levelling (because the Z-axis never sees a reversal), but with auto-levelling it does matter as the Z-axis bobs up and down as the head moves across the bed.

5. if still no joy, did you get a PVP glue-stick with the printer? Try turning the plate over to use the smooth side (cleaned). While it is cold, apply several dabs of glue-stick, and smear it out into a thin film using a slightly damp sponge. Using that side, you can set the bed temperature to reduce to 40degC after the first layer. This will much reduce the tendency of the finished part to shrink and curl as it cools to room temperature.


You are using 80% infill? Generally I have been using 40% for the track bases, and 50% for the filing jigs. The density is likely to affect shrinkage and curling, but by how much I don't know.

I intend to post some suggested Cura profiles for the Neptune 4. I just need a bit more time to experiment with it so that I'm sure I'm not talking rubbish. At present I'm tied up with coding because of an infuriating bug I created in the latest program update.

cheers,

Martin.

 

[James Walters]

@gavin

Great to see your progress.

Based upon my own FDM printing experience, I think Martin has hit the nail on the head re. temperature.

Martin's suggestion of upping the bed temperature and reducing the fans seems the best route forward to me.
It's looking good though. FWIW, on my machine, I found 85degC to be the optimum bed temp and I run with the fan off after the first couple of layers.

Once your settings are dialled-in you'll be well away.

 

[Michael Woods]

Hello Martin, James and others

Experiments with the Neptune 4plus continue over here with some success, but also a few challenges. However I have no doubt it will work out fine - just need to understand the limitations, which is taking a while as I am only doing it occasionally between work on News 237.

I am very much concentrating on being able to print longer lengths - my test is 300mm long. Seems sensible as the purpose of having the larger bed is to enable this.

I seem to have the tops of the sleepers at an acceptable finish now, but now started having problems with the ends of the lengths sometimes curling up part way through the print - I am assuming this is something to do with adhesion on the plate. Whether I have caused this issue by other settings I have not yet worked out.

So I have this time gone back to printing with the Brim (test 6) and will let you know.

The current settings I have changed from the standard profile are:

Global:
Layer Height = 0.12mm
Build Plate adhesion = Brim
Extruder:
Top Surface Skin Layers = 2
Enable Ironing = True
Iron Only Highest Layer = True
Monotonic Ironing Order = True
Ironing Flow = 15%
Infill Density = 80%

No doubt other people have used different settings.

If the picture uploads correctly, here is my curling template corner part way through a print! Off to the 'day job' now so may try post the result of test 6 later.

Gavin
PS Must work out how to get notified for @gavin!

I use an anycubic printer with one of these foil beds.....they work very well but try and position the print in the middle of the plate where you have the max temperature concentration. As you can see the lift is closest to the edge which will be the coolest part of the plate. Plus as Martin and James suggest increase the bed temperature a bit. But if you reposition you may not have to adjust the bed temperature.

 

[Martin Wynne]

@gavin

Hi Gavin,

Here is a test print just made on my Neptune 4. The profile I used is attached below.

There was no warping or curling. You can see at the blue arrow some elephant's foot where the first layer is squidged into the textured bed for adhesion.

Using the latest 244a version of Templot, the finished thickness is 3.3mm after a quick wet rub with the 400-grit sanding block. The design thickness is 3.36mm. Precise measurements are largely meaningless with a heavily-textured bottom surface.

If you are not seeing the elephant's foot, or your thickness is significantly more than 3.3mm, drop the Z-offset a bit, i.e. make it more negative.

One way to judge the quality of the result is to check that the soleplate detail is evident on the toe timber.

The filament is Sunlu PLA+. Generally I prefer the eSun PLA+, but I tried this Sunlu because the colour was supposed to be "wood-like". It isn't.

Here's the Cura profile below if you would like to try it. It's for the Elegoo Cura custom version 5.6. It probably won't import into other versions -- Cura is very picky about accepting custom profiles.

I may change it later of course after more dabbling. There are several changes from the standard 0.12mm profile.

It's for 0.12mm layers, with 0.24mm first layer. Print time for the above 12 timbers was 38 minutes (3.2 minutes per timber).

It's possible to go faster using 0.16mm layers, but I found problems with the soleplate and the dropper retaining ridges. They would need some changes to the Templot default dims.

p.s. after importing it, you must select it in the profile dropdown in order to use it.

cheers,

Martin.

 

[gavin]

Hello all. Thank you very much for the advice above.

Success has been mixed, although progress has been rather delayed by other activities getting in the way. So sorry if it seems intermittent.

I do somehow seem to have gone backwards a bit.

So I took your configuration file, Martin, and James I increased the bed temperature. I also ensured I updated to the latest Templot version.

I then set the printer going and once it was on its way left it running for 20 mins or so then came back to check - by then I had a loose 'golf ball' of plastic wrapped around the print head. It would seem things did not stick properly and all got chewed up. Needless to say, I put it to one side for a week whilst other pressures were dealt with.

So back to it today. Head all cleaned (as far as I can see) but I now seem to not get even the quality I had before. I rolled back all the settings to pre-Martin, pre-James changes, but I am still getting very stringy results which are thus all going wrong.

The picture attached shows the beginnings of the problem. Yes I know it has a brim on - back to my original settings - but you can see that the strings are really messing things up.

I just wondered if anyone else had seen this. At the start of the print it seems to do a neat 'purge' line at the edge of the plate each time but then goes awry! Any advice welcome!

Thanks again
Gavin

 

[Martin Wynne]

@gavin

Hi Gavin,

Try dropping the Z-offset by about 0.1mm to squidge the first layer more firmly into the bed. i.e. after doing the auto levelling according to the book, change the Z-offset (make it more negative), and don't forget to tap the Save button afterwards.

My Neptune 4 is working fine using the profile I posted. Remember after importing it you must actually select it in the dropdown.

Also make sure you have the Neptune 4 printer selected as the active printer -- the Elegoo version of Cura is common across all their printers.

If you now have several gcode files listed in the fluidd jobs list, make sure you right-click on the correct one to start the print. They are not listed with the most recent one at the top. I have now made that mistake 3 times!

cheers,

Martin.

 

[Phil G]

Hi Gavin,

4. have you got the rollers on the vertical columns set too tight, preventing the gravity backlash-correction working on Z? You should be able to lift the X-bar a fraction by hand and it drops back easily under its own weight. As supplied the brass nuts should be slightly loose in the frames. This doesn't matter too much with manual levelling (because the Z-axis never sees a reversal), but with auto-levelling it does matter as the Z-axis bobs up and down as the head moves across the bed.

This quote from Martin is very important, I too had huge problems with jobs curling up, or as you have said coming totally unstuck from the bed. The issue in the end was all about the Z axis binding. My Z wheel positioning as supplied from the Elegoo factory were no good, and I did not pay enough attention to that.
After hours of readjusting I now have all 6 wheels with just a bit of resistance, I also found I had to adjust one of my Z lead screws because as supplied the X frame was racked.
I did this by loosing off one of the timing belt toothed wheels and then manually turning one side until there was no diagonal binding of the Z wheels, IE in my case top left and bottom right on the outside wheels were always very tight. once happy retighten the toothed wheel and re zero and everything was far better. I also found I had to pull the Z vertical columns out a bit at the bottom to ensure they were parallel top to bottom, (this can be done quite easily, because there is a quite a bit of play in the 4 x M5 cap head screws) used to tighten the vertical columns down.
Once all this was done I ran the attached bed leveling STL test. perfect result, when I tried the same test before starting to re-adjust my machine the first square came loose from the bed and was traveling around with the nozzle !!!

You should also check your X axis wheels for excess play, because the Direct drive head is heavy and tends to pull the head down if not set right. You can't see this and it does not matter that much until you try to set your initial Z height with that bit of paper, because that can easily push the head up a bit, if the X wheels are not set right. Again factory setting on my machine were no good at all.
I am not saying yours will be as bad but its worth checking everything is alighted and square because it really does matter.
I hope this helps.
Cheers
Phil,

 

[gavin]

Hello Phil

Thank you very much for all that - it does give me somewhere to look. I think I was quite lucky to start with as I did not seem to need to fiddle with Z-axis adjustment (although that might just have been my naivety!) - however having had the problem with the 'ball' of extruded plastic all around the head I would not be surprised if it has affected the Z-axis adjustment now.

And also that description of how the head moves with the bit of paper - I have started seeing that as well all of a sudden which suggests your analysis is correct.

I will let you know how I get on - will be the middle of the week before I do any more careful testing.

Thanks again
Gavin

PS will try your test print as well

 

[gavin]

@gavin

Hi Gavin,

Try dropping the Z-offset by about 0.1mm to squidge the first layer more firmly into the bed. i.e. after doing the auto levelling according to the book, change the Z-offset (make it more negative), and don't forget to tap the Save button afterwards.

My Neptune 4 is working fine using the profile I posted. Remember after importing it you must actually select it in the dropdown.

Also make sure you have the Neptune 4 printer selected as the active printer -- the Elegoo version of Cura is common across all their printers.

If you now have several gcode files listed in the fluidd jobs list, make sure you right-click on the correct one to start the print. They are not listed with the most recent one at the top. I have now made that mistake 3 times!

cheers,

Martin.

Thank you Martin. I will also work through all of these.

Just to eliminate any possible thing wrong with my software here - I doubt there is but just being certain - could you at some stage send me the file for the simple 12-sleeper sample you printed and I will see if I can get that to print as good here.

No rush
Gavin